WORK PACKAGE 2
Safety Impact assessment and cost-economic assessment in ASSESS

One of the goals of the ASSESS project is to increase consumer awareness of the functionalities and benefits of forward-looking collision mitigation systems for passenger cars. Work package 2 aims especially at providing figures of the benefits of the systems in monetary terms. The core of benefit assessment is the estimation of the safety impacts. The specific steps of this impact estimation are firstly shown before the key aspects of the methodology for mitigation estimation are described. Finally different variants of monetary assessment are presented in which the test-based impact assessment are implemented.

 

Steps of impact assessment

Socio-economic evaluation of Pre-crash warning and autonomous braking systems in ASSESS is test-based, i.e. the outcomes of pre-crash and crash tests are used in a systematic way to estimate effectiveness of the safety systems. Figure 1 shows the different methodological steps which are necessary to incorporate test results in safety impact assessment:

 

 

Figure 1: Safety impact “logic” for track test scenario i (own figure)

 

The tests are undertaken for different scenarios specifying driving speed of subject and target vehicle, behaviour of target vehicle (braking, stopping), offset of vehicles in case of crash, driver reaction and so on. For practical reasons and because of feasibility only a limited number of scenarios and tests will be undertaken. Therefore, the first step of impact estimation is to check whether the test is feasible. If a test is feasible (1) then depending on whether the test results in complete crash avoidance or in crash mitigation different impact estimation methodologies are applied. For the case of avoidance (2) the safety impact addresses the complete target population of the scenario i (TPi), thus system effectiveness is 100%. The impact assessment methodology then mainly has to provide weighting factors which describe the share of the target population of rear-end collisions which the test scenario represents. For the case of mitigation a change of collision speed results, and therefore a share of the serious injuries of the target population can be prevented. (3). Of course, accident consequences depend on several additional factors (e.g. relative car masses, age and gender of occupants, offset), but impact speed (transferred to delta-v) will also have a strong influence on accident consequences. The mitigation case demands a more complex methodology for impact estimation compared to the avoidance case which will be explained below. So far only feasible tests were considered. Also a strategy for non-feasible tests (e.g. no driver reaction, impact speed e.g. above 80 kph) has to be provided (4) for a complete impact estimation. This could entail an extrapolation of test results of similar scenarios to the non-feasible test done by experts, or assuming no impact if no such similar scenario is available.

 

The mitigation model

The mitigation model is based on S-shaped injury risk curves formalising the relationship between casualty of different seriousness and delta-v. The curves show the risk change in a crash situation for a car occupant if delta-v is reduced. The curves used are based on GIDAS data including all collision types, and belted occupants drivers are assumed (HANNAWALD 2008). The curves show the risk change for a car occupant if delta-v is reduced in a crash situation:

 

 

Figure 2: Injury-risk of passenger car occupants depending on delta-v for injury level fatal, serious and slight injury (based on HANNAWALD, L., 2008)

 

Injury-risk curves specify injury risk of occupants per accident. Thus, for estimation of the safety impact over all possible accidents, the risk change in a given crash situation has to be combined with the frequency and hence the distribution of rear-end collisions with respect to delta-v. This distribution describes the share of accidents which hypothetical can be addressed by speed reduction within the accident target population. In the model a Lognormal distribution of accidents over impact speed parameter delta-v is used following US-accident data because similar data at this level of detail are missing for Europe. By use of simulations a hypothetical shift of the accident distribution function caused by a reduction in impact speed is produced as shown in the figure below. Combined with injury-risk functions then percentage reductions of casualties are predicted. These changes are interpreted as system effectiveness with respect to different reductions of delta-v caused by the pre-crash system. Of course, if necessary effectiveness has to be weighted with regard to performance limits of the systems e.g. detrimental weather conditions.

 

 

Figure 3: Shift of accident frequency of over delta-v (based on KONONEN, D., et al., 2010)

 

The derived effectiveness figures are then applied to the accident target population to get the maximal reduction potential of casualties. (See intermediate deliverable D2.2 (1/2) of ASSESS for the procedure of forecasting the accident target population for EU-27 for the assessment horizon 2020 and 2030.) For getting a more realistic impression of the magnitude of the safety impacts they have to be weighted with the expected fitment rate of the car fleet with the systems.

 

Approaches of test-based cost-benefit analysis in ASSESS

In cost-benefit analysis in road safety costs and benefits of safety measures are compared for calculation of Benefit-cost ratios (BCRs). However, since a reliable estimation of system cost actually is not possible instead of calculation benefit-cost ratios critical safety system costs are calculated by Break-Even Analysis (BEA). The figure below shows the different ways of BCRs calculation which will be run in ASSESS for evaluation of pre-crash warning and autonomous braking systems:

 

 

 Figure 4: Variants of cost-benefit analysis (own figure)

 

Socio-economic assessment of road safety measures is done based on break-even analysis of critical system costs (BEA). Critical system costs are defined by equating aggregated safety benefits of vehicles with the system on board and the corresponding fleet costs.

Three different approaches are chosen which differ with regard to their assessment goal, but all variants use ASSESS testing outcomes for estimation of safety benefits:

Following eIMPACT project, and the safety impacts derived there by experts´ safety impact estimation, the pre-crash systems may avoid and mitigate accidents of accident types other than rear-end collisions which are in focus of testing in ASSESS (e.g. single vehicle collisions with obstacles beside road, angle collisions). These additional safety impacts are included in impact estimation.

• Test-based assessment
• By testing the system’s performance under realistic conditions, the ASSESS project is able to show evidence that the focused systems are able to avoid or mitigate crashes according to the test scenarios based on accident analysis. Thus compared to expert and theoretical based impact estimation the test results provide first empirical evidence for estimation of the safety potential of the systems. Hence, safety benefits of the systems are evaluated as accurate as possible given available empirical evidence provided by ASSESS project.
• Ranking of different functionalities of pre-crash systems based on effectiveness and efficiency estimation: Testing of the systems can lead to identification of performance differences of systems of different functionalities. However, the figures of safety impact of different systems will reflect differences of generic safety systems. The calculated critical system costs can be used to rank pre-crash systems of different functionalities (warning-only system, pre-crash system comprising warning and autonomous emergency braking etc).